We previously showed that hormonal factors such as early menopause and breastfeeding history influence the risk of RA [2
]. In particular, early menopause was a robust risk factor for RA [14
]. In this study, we demonstrated that early menopause predicts a milder type of RA.
Estrogen is suggested to suppress cellular immunity directly but to stimulate the humoral immune system [21
]. A decrease in estrogen levels may thus contribute to a T-cell differentiation skewed toward the T-helper 1 (Th1) subset. This has previously been implicated as an important part of the pathogenesis of RA [22
]. Further studies should examine the impact of the timing of menopause on immunologic and inflammatory pathways leading to RA.
Dysfunctions of the hypothalamic-pituitary-adrenocortical axis (HPA axis), the hypothalamic-pituitary-gonadal (HPG axis), and the autonomic nervous system have all been suggested to be a part of the complex pathogenesis of RA [24
]. Similar observations have been made in primary Sjögren syndrome [26
]. A gradual decline in the function of the HPG axis is considered a key element in the development of menopause [27
]. We propose that women with a less-responsive HPG axis, leading to an increased risk of early menopause, may also have a primarily malfunctioning or "sluggish" HPA axis. A reduced response of the HPA axis to inflammation would make them more susceptible to develop a chronic inflammatory disease. In support of this, it has been found that women who develop RA later in life have higher birth weight than do normal controls [29
], and some data link high birth weight with a less-responsive HPA function later in life [31
]. This may reflect an impact of HPA function on chronic inflammation in general, rather than on the specific pathogenesis of erosive arthritis.
In this study, women with a history of early menopause in whom RA developed were more likely to have a mild disease course, and less likely to progress to severe RA. This suggests that different predictors may be associated with distinct clinical outcomes. For example, other known risk factors for RA, such as the shared epitope of HLA-DRB1, and smoking, may, in comparison with early menopause, be more specifically associated with a severe phenotype of RA [32
A recent study [12
] suggested that OC use is associated with mild RA and demonstrated a reduction of disability, measured by HAQ over time, in former OC users, compared with nonusers. Furthermore, current OC users had lower HAQ scores at baseline and over time than did previous users. Jorgensen et al.
] found that OC is both protective for RA and associated with a mild phenotype in individuals who develop RA. Yet another study did not find a protective effect of current use of OC for RA [34
]. Because our patients were older, with a mean age at diagnosis of 63 years (range, 47 to 80 years), only one patient reported current use of OC at diagnosis, and we could therefore not address the effect of current use of OC. The mean time between OC use and RA diagnosis was 20 years, which may explain the limited impact of former OC use on the severity of RA. In our sample, the former users rather tended to develop a more-severe RA, at least by some outcome measures. This could be biased by their younger age at diagnosis. In particular, the difference in the proportion treated with biologics may be due to more extensive prescription of such drugs for younger individuals with RA, as was the case in the present study and also reported in a recent national survey [35
Barrett et al.
] suggested that breastfeeding is a risk factor for a more-severe inflammatory polyarthritis in a small group of genetically susceptible women, based on the linkage between HLA-DRB1 alleles and the prolactin gene on chromosome 6 [37
]. In a prospective short-term study of pregnant women with inflammatory polyarthritis, they found that women who breastfed after a first pregnancy had a more-severe disease 6 months postpartum compared with non-breastfeeders and previous breastfeeders [36
]. However, in the present study, we could not demonstrate any significant long-term effect of breastfeeding on the severity of RA. Our study design excludes the smaller group of women who develop RA during their younger years, which could affect our results. The immune-stimulating effect of prolactin could also be short term, and breastfeeding could have a different long-term immune-modulating effect, suggested by the results from the Nurses Health Study [38
] and our previous results [2
Limitations of this study include the retrospective data collection, because information in medical records is sometimes incomplete and not always coherent, due to diverse assessments by different physicians. Because risk-factor information and clinical records are two completely unrelated sources of information, it is unlikely that such incompleteness is differential between risk factor-exposure categories. Nevertheless, the relatively large proportion of missing data of HAQ at specific time points is a limitation. Because some physicians may be less prone to collect HAQ routinely, and this may also change over time, we cannot exclude a systematic bias based on this. However, no major differences were found in the proportions with missing HAQ between those with early and normal/late menopause.
Because of differences in management practice, the methods used to define clusters of disease severity may not be valid in other populations. Another limitation is the relatively small sample size, which affects analyses of subsets with different risk-factor profiles.
Strengths include the prospective assessment of hormone-related factors before disease onset, which limits the risk of recall bias, and the population-based approach, which reduces the risk of selection bias.